【摘要】 化工过程的回路众多、关系复杂,一个回路的振荡可能会波及其他回路,多个回路的异常往往只由一个故障点引起。因此分析多回路的振荡传播路径并找到故障根源对于消除故障十分重要。流程工业普遍存在的低频噪声会干扰一般振荡检测方法的准确性,本文以TE过程为例,用频域滤波后的自相关函数(ACF)解决低频干扰严重且频率非单一的振荡检测问题,通过对多回路过程输出的功率谱提取主元,并根据主元相似性对回路聚类,分析振荡的传播路径,最后借助双相干等非线性检测方法定位厂级振荡的根源。同时本文基于上述方法开发了振荡检测与诊断工具箱,对多回路系统的数据进行自动分析。更多还原

【Abstract】 In chemical industry hundreds of control loops connect to each other by feedback or cascading mechanism.One loop’s oscillation may affect the downstream loops and multi-loop oscillation alarms usually come from the same disturbance.Localization the source of oscillation is most important for removes its impact.Low frequency noises may deteriorate the performance of oscillation detection methods. By removing low frequency,Auto-covariance function is employed to detect oscillation in TE problem.Spectral principal component analysis and nonlinearity detection are utilized to cluster the loops with similar problem and localize the oscillation root cause.In the mean time,a tool box integrated the above methods is developed to provide automated results of multi-loop data.更多还原